Документ взят из кэша поисковой машины. Адрес оригинального документа : http://xmm.vilspa.esa.es/calibration/documentation/epic_cal_meetings/201504/Ness_Cal_2015.pdf Дата изменения: Fri Mar 27 12:39:00 2015 Дата индексирования: Wed Apr 13 13:28:12 2016 Кодировка: Поисковые слова: mercury

EPIC-pn Burst Mode X-ray loading and rate-dependent CTI calibration
Jan-Uwe Ness ESAC 09 April 2015

XRL/RDCTI corrections in burst mode

JU Ness

ESAC, April 9 2015


Energy offsets caused by X-ray loading

In 2010, significant effects from X-ray loading (XRL) were discovered in the timing mode and burst mode offset maps (Guainazzi & Smith 2013, XMM-CAL-SRN-0302) => As of 23rd March 2012 offset maps with closed filter rather than science filter But pre-2012 observations need XRL correction Causes not understood => Empirical correction: PHA = a + b в XRL In Timing Mode a=0, b=1 already implemented Since in burst mode, no offset at low count rates: a=0. Further b=1 was chosen and subsequent Rate Dependent Charge Transfer Inefficieny (RDCTI) calibration adjusted:

XRL/RDCTI corrections in burst mode

JU Ness

ESAC, April 9 2015


Rate Dependent CTI correction

Energy shifts caused by Charge Transfer Inefficiencies (CTI) are correced with a parameterized function (hard-coded in the SAS) called Gain factor: G=(a0 Ca1) + a2 with C the average number of shifted electrons in the events file and a0, a1, a2 adjustable parameters that are stored in the CCF component EPN_CTI_0045.CCF where 0045 is the current version number. Thus the values of a0, a1, a2 control the rate dependent correction and are determined empirically.

XRL/RDCTI corrections in burst mode

JU Ness

ESAC, April 9 2015


Energy scale correction: Determine energy shifts

To detect possible energy shifts in the final spectra, distinct features such as emission lines would be ideal, but most burst mode spectra only contain continuum with very weak emission/absorption lines which might even be shifted by high velocities Use instrumental edges of Si (CCD) and Au (mirror) that are included in the response Method: perform a spectral fit including a gain parameter that shifts the entire spectrum Ideal: gain parameter zero Use non-zero gain parameters to empirically determine calibration parameters

XRL/RDCTI corrections in burst mode

JU Ness

ESAC, April 9 2015


Energy scale correction: Sample

XMM-Newton Science Archive (XSA) contains 123 burst mode exposures (small sample) Since we are calibrating rate dependent effects, only observations of non-variable sources can be used. Only include sources with variations within 3 times the standard deviation => even smaller sample Sources with sharply-peaked variability patterns can be included when applying time filtering (gti)

XRL/RDCTI corrections in burst mode

JU Ness

ESAC, April 9 2015


Energy scale correction: Current status

The currently public CCF for burst mode contains energy scale corrections · XRL: a=0, b=0 EPN_REJECT_0007.CCF thus no correction · RDCTI: a0=0.003, a1=0.3050, a2=0.943

withdefaultcal=yes

XRL/RDCTI corrections in burst mode

JU Ness

ESAC, April 9 2015


Energy scale correction: Without any corrections

The currently public CCF for burst mode contains energy scale corrections · XRL: a=0, b=0 EPN_REJECT_0007.CCF thus no correction · RDCTI: a0=0.003, a1=0.3050, a2=0.943

withdefaultcal=NOoooooo no XRL: a=0, b=0 no RDCTI: a0=a1=0, a2=1 => G=1

XRL/RDCTI corrections in burst mode

JU Ness

ESAC, April 9 2015


Energy scale correction: Only XRL correction

The currently public CCF for burst mode contains energy scale corrections · XRL: a=0, b=0 EPN_REJECT_0007.CCF thus no correction · RDCTI: a0=0.003, a1=0.3050, a2=0.943

New: XRL: a=0, b=1 => recalibrate a0, a1, a2 for RDCTI

XRL/RDCTI corrections in burst mode

JU Ness

ESAC, April 9 2015


Energy scale correction: Determine parameters for gain factor
Normalise gain shifts from spectral fits to the energy of the gold edge: G=gain/2200eV+1

Perform a fit to datapoints with a0, a1, a2 as free parameters

XRL/RDCTI corrections in burst mode

JU Ness

ESAC, April 9 2015


Energy scale correction: Current status

The currently public CCF for burst mode contains energy scale corrections · XRL: a=0, b=0 EPN_REJECT_0007.CCF thus no correction · RDCTI: a0=0.003, a1=0.3050, a2=0.943

New: XRL: a=0, b=1 RDCTI: a0=0.1, a1=0.047, a2=0.8

XRL/RDCTI corrections in burst mode

JU Ness

ESAC, April 9 2015


Energy scale correction: Validation

One burst mode exposure contains emission lines, Cas A, and there is a Large Window exposure for direct (model-independent) comparison:

The new calibration (red) is slightly worse than present one (blue) But: this is the faintest target in the sample and not representative for typical use of burst mode

XRL/RDCTI corrections in burst mode

JU Ness

ESAC, April 9 2015


Energy scale correction: Validation

GRO J 1655-50 has absorption lines of Fe XXV

The new calibration (red) is clearly better than present one (blue)

XRL/RDCTI corrections in burst mode

JU Ness

ESAC, April 9 2015


Energy scale correction: Validation

4U 1700-37 has Fe XXV/XXVI emission lines and a neutral Fe absorption edge:

The new calibration (red) is much better than present one (blue) Note that these observations have no offset map, thus XRL correction not effective

XRL/RDCTI corrections in burst mode

JU Ness

ESAC, April 9 2015